Flying a helicopter low and fast over rugged terrain, using the hills for masking, to a safe landing in a forest clearing is among the most demanding tasks a pilot can undertake, but one U.S. Army researchers believe can be performed autonomously - with the right sensors and algorithms.
Under the Autonomous Rotorcraft project, the Army’s Aeroflightdynamics Directorate (AFDD) has developed and flown software that would enable “terrain-aware autonomy” in the next generation of unmanned rotorcraft.
In Phase 2 flights in May, using a laser-radar sensor, the software guided a fly-by-wire testbed helicopter through the Diablo Range west of San Jose, Calif., autonomously navigating out of a box canyon, avoiding an aircraft and power lines, and selecting a safe landing area.
AFDD has posted videos from the autonomous flights on Youtube. They show the view from the helicopter and, inset left, the terrain display generated by the 3D-LZ scanning lidar (visible at top right of the picture).
The first shows the JUH-60A RASCAL testbed finding its way out of a box canyon using the lidar and an algorithm called Obstacle Field Navigation (OFN):
The second shows the lidar detecting power wires in the helicopter's path and the OFN algorithm commanding the RASCAL to climb over them:
The third was accidental, it seems, and shows the system detecting the chase aircraft when it strayed into the RASCAL's path, and OFN commanding the helicopter to slow down and steer round the "obstacle":
The fourth shows the second algorithm developed by AFDD, Safe Landing Area Detection (SLAD), in action - the lidar checking the landing zone for slope, surface and obstacles and selecting a safe touchdown point:
All videos: US Army
AFDD plans more flights later this year with a longer-range lidar. The directorate says the RASCAL's autonomous flight behavior is so good it can be flown unmanned faster than expected, and the 600m-range 3D-LZ lidar can't see far enough ahead to select a safe landing area in time. The new lidar will have 2km range.